Friday, March 3, 2017

Playing up - Is it good or bad for USTA League play? Interesting Tennis League Stats

[Warning: This is a longer blog post with lots of statistics, but please do read through to the end!]

USTA League play is intended to encourage matches to be played by players of similar ability and is accomplished through the use of the NTRP rating system.  This assigns players a level based on match results, and then player join teams in leagues at their designated level.  While there are occasional exceptions and strange things that come out of the system, by and large, it works and most matches are competitive.

For example, in both the 18 & over and 40 & over leagues, a full 26% of matches go to a third set.  And if you define competitive as the losing player/pair winning at least four games, a full 80% are competitive.  If you go by a stricter definition of at least six games, the percentage drops to 60%, but still, for the most part, league matches are competitive.  And yes, at least five games is right in the middle at 70%.

But what about the non-competitive matches, the 20% where the losing player/pair win three or fewer games?  First, there are very few double bagels, less than 2% of matches end with that score, and barely over 4% are 1 & 0 scores.  That leaves 15% where the losing pair win two or three games.

Here is a chart showing the percentage of time the losing player/pair loses each number of games.

So, what can we see from the 20% of the matches that aren't competitive?  Using my Estimated Dynamic NTRP Ratings and looking at the difference in ratings between the winning and losing player/pair in these matches, we see it averages 0.23, a pretty big gap in favor of the winning player/pair as you might expect.  Correspondingly the difference in the competitive matches averages just 0.08, the player/pair that wins is still favored, but by a lot less.

An average rating difference of 0.23 can be achieved a few ways.  That is about half the range for a level, so can easily happen when a high 3.5, say 3.4, plays a low 3.5, say a 3.17.  But being an average, that means it is likely the rating difference is quite a bit larger at times, and it becomes more difficult for two at-level players to be playing in this case.  This means that at times, the difference is due to players playing up.

Digging deeper and looking at how many times players are playing up in non-competitive matches, we see that in singles, 28% of non-competitive matches have the loser playing up against an at-level player.  Doubles is worse with 25% of matches having one lower playing up against two at-level players, and 14% of the non-competitive matches having both losers playing up against two at-level players, meaning a full 39% of non-competitive doubles matches have at least one player playing up against at-level players.

It is not a surprise that players playing up can result in non-competitive matches, but when you see the stats, it appears one could eliminate nearly 40% of the uncompetitive matches if players weren't playing up.

Which raises the question, is allowing players to play up good or bad for USTA League tennis?

On one hand, allowing playing up accomplishes a few things:
  • Teams with a small roster of at-level players can be supplemented to have a large enough roster to ensure all five courts can be fielded each match.  Or without players playing up, at-level players may not be able to field a team as they just don't have enough.
  • Players right at the threshold of being bumped up can challenge themselves against stronger players to help them try to improve their game.
  • When players right at the threshold play up and get matched up against lower rated at-level players, it can be a competitive match between players with ratings just 0.1-0.2 apart.

But on the other hand, there are some negatives:
  • Those playing up are often over-matched leading to uncompetitive matches.  This can happen especially when the playing up player is sacrificed on court 1 to try and allow their team to win courts 2 and 3.  And since anyone can play up, you could have a 2.51 "3.0" playing up against a 3.50 "3.5", a full point apart.  This goes against the purpose of the NTRP system and league tennis, to have competitive matches, and can be considered a waste of time for the at-level player to have to take the time and expense to drive to perhaps an away match and perhaps pay court fees to play a "joke" of a match.
  • Ratings can easily be skewed as the NTRP system doesn't work well or take into account some of the unavoidable by products of matches between players rated far apart.  For example, if a strong at-level player plays a weak below level player, the at-level player may be expected to not give up a game.  If they simply get lazy or purposely lose some games, or just want to be nice and not dish out the double bagel, their rating will drop and the below level player's rating will go up inappropriately.  If this happens several times, player's ratings can move enough that an unjustified bump up or down can occur.

Unfortunately for those that see the negatives as a big issue, USTA regulations don't put many limitations on playing up.  They allow up to 50% of a roster to be below level players.  And there are no restrictions on how many can be in a given line-up.  This means it is possible for an entire line-up to be below level players.

Now some sections do adopt their own rules that limit the below level players to a smaller percentage, but there are also provisions by National that all sections adopt that allow for more than 50% to be below level if the team went to Nationals the prior year and moves to a higher level flight as a team.

If one were to want to allow playing up, but try to address some of the negatives, a few things could be done:
  • Playing up could be outlawed so it isn't allowed at all.  This likely isn't going to happen.
  • The number of below level players in a line-up could be limited and perhaps be just two or three.  This would limit how often players with a large ratings discrepancy face each other.  This seams reasonable.
  • The plus leagues (40+ 4.5+ and 18+ 5.0+) have specific rules about how many above level players can be rostered (3 or 4) and play in a match (2), and also what court they must play on (1 singles or doubles), so something similar could be done for players playing up.  Limit the number to say 2, and mandate that they must play on 2 singles or 3 doubles.  This would preclude using low rated below level players as sacrificial lambs on court 1.  This also seems reasonable.
  • Introduce restrictions on who can play up, for example require that a player's year-end rating have been in at  least the upper half or even 25% of their level to preclude the lower rated below level players from playing up.  This would mean the USTA would have to disclose this information, and they tend to like to keep it all a secret, but in Georgia they have regular and "low" leagues which effectively is disclosing the same information.  So perhaps this could be done.

I'll be honest, I'm torn on the subject.  I do tend to lean towards limiting playing up and/or having some restrictions.  But I actually fall into the camp of captaining a team that needed players playing up to field a line-up and while I always played my available and healthy at-level players over below level players, I still had to run a few below level ones out every match, although it was usually just one or two and was three only once.

But I would probably support the third bullet above, adopt something similar to the plus leagues for players playing up.  That still allows it, but limits the negatives.

What do you think?  Should players be allowed to play up?  If so, should there be some restrictions of some sort?